GB1417741A - Methods of conditioning the operation of data processing apparatus - Google Patents

Abstract

1417741 Data processing SCHLUMBERGER INLAND SERVICES Inc 24 Jan 1973 [19 May 1972] 3604/73 Heading G4A A curve represented by a series of signal samples is analysed in a data processing machine to identify features such as peaks, depressions and surges and typical combinations thereof, the machine storing the value of at least one abscissa and a signal indicating the type of feature for each feature identified from the signal samples, and computing and storing a plurality of parameters representing the form of each identified feature. The information so obtained may be used in correlating a number of curves derived, for example, from well logging devices or seismic surveying equipment. Curve slope calculation.-For each sample point i the differences between pairs of samples located on either side of i are calculated and weighted according to a linear law decreasing with distance from i, the weighted values being summed. Identification of peaks.-Adjacent pairs of slope values are compared until the new value (i+1) becomes smaller than the old (i), then successive new values (i+2) and so on are compared with the value for point i until a threshold S is exceeded by the difference in slopes. Several different threshold values Sj may be applied to define the size of the peak. Should the slope at a point (i+r) exceed the slope at point i before the threshold is exceeded, a peak is not identified and the search continues by comparison of the slopes for (i+r) and (i+r+1). Further, a peak is identified at a point i only if the curve increases and decreases continuously beyond a threshold, on either side of the point. The boundaries of a peak are taken as the samples corresponding to the maximum and minimum slope which flank the peak. Identifications of depressions is broadly similar to the procedure for identifying peaks. Identification of surges.-A surge is defined by the existence, over a distance D to one side of a point i, of a series of maximum levels at or below a series of minimum levels over the same distance D to the other side of point i. In such a case the average of the lowest minimum and highest maximum is found and the closest sample to this average value is taken as the point defining the boundary of the surge. Alternatively, a surge may be defined by the existence of a difference between the average values A, B of samples on either side of the point i which exceeds a threshold, the boundary then being taken as the sample closest in value to the average of A and B. Form parameters.-For a peak or depression these include length, position of its maximum or minimum, its value and slope at its two ends, and its average value, and for a peak, its fineness, height and ratio of the lengths of its flanks. The parameters for a surge include its value, sign and the flank slope. Typical combinations of features are identified from the identified individual features, their relative positions and their calculated parameters. Each such combination identified is identified by type and stored together with relevant position and parameter data. Correlation of curves.-Initially correlations between the most outstanding features, i.e. those with exceptional form parameter values, are searched for. The search is, in addition only carried out among features of the same type or of compatible types and among features which are likely to correspond, taking into account any correlations previously established. For each feature on a curve C, a set of possible corresponding features on another curve C<SP>1</SP> is derived from examination of the relative positions of the respective boundaries. From the corresponding form parameters p r , p r <SP>1</SP> of the point and each possible corresponding point, a correlation coefficient is calculated as the sum of the squares of the differences between the various p r , p r <SP>1</SP>. If the difference between the two smallest coefficients so obtained does not exceed a first threshold, there is ambiguity and correspondence is not established. If the first threshold is exceeded, the lowest coefficient is compared with a second threshold and if this threshold is not exceeded, correspondence is established. If there is only one possible corresponding feature, the correlation coefficient indicates correspondence if it is less than the second threshold. Establishment of a correlation between two features may necessitate changing the boundary values assigned to other features.